Detecting hot pixels

• We had two 1 hour exposure dark frames. One was at an angle and the other pointed upwards.
• We wanted to find the x and y coordinates of hot pixels and the hotness of those pixels for each image.
• Wrote a for loop program to do that.
• Wanted to investigate if the hot pixels were clustered for each image.
• To do that we plotted a histogram of the frequency of hot pixels in x and y coordinates for each image.
• For the upwards picture 6951 pixels were found for the threshold of 2000.
• For the image at an angle 5748 pixels were found for the threshold of 2000.
• In total there are 1560600 pixels.

Histograms for x and y positions

• x coordinate histograms for upwards and at an angle.

• y coordinate histograms for upwards and at an angle.

• The plots show where there are more hot pixels. In this case it is at the edges of the CCD.
• Plots agree for both images.
• For upwards image there are more hot pixels and it would explain why there are more pixels for a given x or y bin.

Histograms for hotness of pixels

• The hotness of the hot pixels was plotted as a histrogram for both images.

• For both images the shape of the histogram is similar.
• There is an unusual spike.
• At an angle the spike is at 4000 pixel values, whereas when the camera is pointed upwards, the spike appears at 6000 pixel values.

Comparing hotness of the two independent 1 hour exposures

• We wanted to investigate how the hotness changes with the angle of the camera.
• We had two images, upwards and at an angle.
• We matched the hot pixels between the two lists.
• Some pixels were matched more than once, because some pixels had no corresponding hot pixel in the image, this is because the upward image had a lot more hot pixels.
• The correspond histograms are shown below for different threshold pixel values.

• Due to computing time many different threshold values were used.
• We could only go down to 3000 pixel values as the threshold.
• We expected a Gaussian centered on 0, because we expect a symmetric difference with the most common difference being zero, and the difference is less likely.
• As can be seen, especially at large thresholds the shape deviates from a Gaussian and it is not centered on 0.
• Some differences are very high.
• For the threshold of 3000 pixel values, there were 5573 hot pixels for upwards image and 4453 hot pixels in the angle image.
• It appears that there are always more hot pixels in the upwards image as compared to the image at an angle.